PSI - Issue 33

Umberto De Maio et al. / Procedia Structural Integrity 33 (2021) 954–965 Author name / Structural Integrity Procedia 00 (2019) 000 – 000

963

10

errors of the crack propagation angle, due to an inaccurate prediction of the principal stress direction induced by the adopted mapped mesh which introduces preferential mode-I crack directions.

Fig. 8. Comparison between the numerically predicted main crack path obtained by plotting the damage variable along the interface and within the solid elements for DIM and ECM, respectively. 4. Conclusions In this work, two different finite element-based cohesive fracture models to simulate cracking phenomena in concrete materials, are compared. The first one is a novel diffuse interface model (DIM), based on the insertion of cohesive interface elements along all the internal mesh boundaries within a standard displacement-based finite element setting, capable to simulate multiple crack nucleation and propagation under mixed-mode loading conditions in a very accurate and computationally efficient manner without any additional insertion criterion and/or complicated remeshing. The second one is the embedded crack model (ECM) which relies on the strong discontinuity approach, able to introduce the displacement jump in the kinematics of the corresponding finite element, and on a cohesive zone model, employed to simulate the cracking behavior. In this way, multiple crack propagation along nonprescribed paths in quasi-brittle materials is successfully simulated without requiring sophisticated crack tracking and mesh update procedures. Several simulations of cracking behavior in plain concrete specimens subjected to mode-I and mixed-mode loading conditions have been performed by using the DIM and ECM models. The numerically predicted results, in terms of load carrying capacity and crack patterns, are globally in good accordance with that obtained by experiments and numerical models taken for comparison purposes. In detail, the DIM predicts a stronger structural response due to the artificial toughening effect induced by tortuosity of the main crack, leading to a little peak load overestimation, respect to that obtained by the reference. Contrarily, the predictions of the ECM appear to be almost overlapped to the reference curves in both peak and post-peak stage, resulting in a negligible peak load relative error with respect to the reference. The crack patterns predicted by both models are very consistent with the available experimental outcomes. However, the adoption of a diffuse interface model dramatically simplifies the prediction of complex fracture phenomena in concrete elements allowing single crack paths and/or patterns to be obtained in a more realistic manner than the ECM approach. In conclusion, the numerical outcomes have confirmed the reliability and effectiveness of both models to simulate cracking phenomena in quasi-brittle materials. However, the ECM approach provides numerical predictions, in terms of load-carrying capacity, very consistent with the experiments, but requires in-depth programming skills respect to the DIM approach, as changes of existing FEM codes to condense the additional degrees of freedom in the cracked elements displacement field are needed. On the other hand, the DIM approach relies on classical interface formulations easily implementable in the most used commercial finite element codes, thus dramatically simplifying the solution of fracture problems and allowing the related crack patterns to be predicted in a realistic manner.